There was a quarter — I won’t say exactly when, but it was in our second year — where everything came apart at once. We lost two clients in the same month. A key team member gave notice. A product milestone we’d been working toward for months slipped by six weeks. Our cash position, which had always been comfortable, suddenly wasn’t.
None of these events were individually catastrophic. Together, they created a compounding crisis that tested Bruno and me more than anything before or since. And the uncomfortable truth is that almost all of it was preventable. The systems we should have had in place — the ones we built after — would have caught the warning signs months earlier.
This is the story of that quarter and what it taught us.
The Client Losses
The two clients we lost had something in common: both relationships had been deteriorating for weeks before the formal churn, and we hadn’t seen it.
One client had been growing increasingly unresponsive to our deliverables. Turnaround times on their approvals stretched from days to weeks. Their founder stopped attending our weekly syncs. We interpreted this as them being busy — which they were — but busy was a symptom, not the condition. The condition was that they’d lost confidence in the engagement. They didn’t believe the work was moving the needle, and instead of telling us, they quietly disengaged.
The other client’s departure was more straightforward. They’d hired an in-house marketing lead, and the new hire — reasonably, from their perspective — wanted to bring the work internal. We knew about the hire. We didn’t anticipate what it would mean for the engagement. In retrospect, the moment a client hires a senior marketing person, the agency relationship is at risk. We should have had a conversation about how the engagement would evolve, not waited for the conversation to be forced on us.
Both losses hit our revenue hard. Losing one client is manageable. Losing two in the same month is a structural problem, because client acquisition has a lead time. You can’t replace two clients overnight. The pipeline we had wasn’t deep enough to absorb the shock.
The Talent Loss
The team member who left wasn’t disgruntled. They were talented, they’d been doing excellent work, and they got a better offer from a company that could pay more than we could at the time. That’s fair. I don’t begrudge anyone for taking a better opportunity.
But the departure exposed something we hadn’t addressed: single points of failure in our delivery team. This person was the primary operator on three client accounts. They held the relationship context, the institutional knowledge of each client’s preferences and history, and the muscle memory of the recurring workflows. When they left, that knowledge walked out with them.
The transition was rough. Deliverable quality dipped for about three weeks while we scrambled to redistribute the work and get the new operators up to speed. Two of the three affected clients noticed. One of them — I’ll be honest — was one of the two we later lost.
The Cash Crunch
Our financial model had always assumed steady revenue and gradual growth. It didn’t account for what happens when you lose 25% of your revenue in thirty days while carrying fixed costs — salaries, tools, contractors — that don’t flex downward at the same speed.
We weren’t in danger of going under. But we went from comfortable to tight, and tight changes everything. It changes how you think about investments. It changes the tenor of conversations about the product roadmap. It changes the anxiety level that you bring to every decision.
Bruno and I had our most difficult conversation during this period. Not a fight — we’ve never had those — but a genuinely tense discussion about whether we should pause the product development to shore up the agency, or keep investing in the product and bet that the agency would recover. We chose a middle path: we slowed the product timeline and redirected some of that energy toward client acquisition. It was the right call, but it cost us three months on the product roadmap.
What We Built After
The quarter forced us to build systems that we should have built from the beginning. Here’s what changed.
Client health scoring. We implemented a simple but rigorous system for tracking the health of every client relationship. It measures four signals: engagement level (are they responsive, do they attend meetings, do they give timely feedback), deliverable satisfaction (are revisions increasing, are they pushing back more than usual), strategic alignment (do they still believe in the direction we’re heading), and champion stability (is our internal champion still in place, is their authority intact). Each signal gets rated monthly. Any account that drops below a threshold triggers a proactive conversation.
Since implementing this system, we haven’t been surprised by a churn. We’ve still lost clients — that’s inevitable in agency work — but we’ve seen every one coming and had time to either address the issue or prepare for the transition.
Knowledge documentation. Every client account now has a living document that captures the relationship context: who the stakeholders are, what their preferences are, what the recurring workflows look like, what the history of key decisions is. When someone transitions off an account, the incoming operator has a comprehensive briefing rather than a scattered collection of Slack messages and tribal knowledge.
Financial stress testing. Instead of modeling for steady-state growth, we now model for scenarios. What happens if we lose our largest client tomorrow? What happens if two clients churn in the same quarter? What happens if a major expense comes in higher than expected? Each scenario has a predetermined response — cut this, defer that, accelerate the other thing. The response plans are reviewed quarterly.
Pipeline depth targets. We set a minimum pipeline depth — the amount of active opportunities at any given time — and treat it as non-negotiable. When the pipeline drops below the target, business development becomes the top priority regardless of what else is happening. This ensures that we’re never in the position of losing a client and having nothing in the funnel to replace them.
The Lesson
The worst quarter taught me something I intellectually knew but hadn’t internalized: the systems you need most are the ones that feel unnecessary when things are going well.
During our good quarters — and we’d had several in a row before the bad one — building a client health scoring system felt like overhead. Documenting tribal knowledge felt like bureaucracy. Financial stress testing felt like pessimism. We were growing. Clients were happy. The team was stable. Why build for a scenario that wasn’t happening?
Because by the time the scenario happens, it’s too late to build the system. You’re in reactive mode, fighting fires, making decisions under pressure with incomplete information. The systems that save you in a crisis are the ones you build in calm.
I think about this constantly now. Whenever I’m tempted to defer a process improvement because “things are fine,” I remember the quarter when things weren’t fine and the process improvement didn’t exist yet.
Things are always fine until they aren’t. Build the system while they’re fine.